System for identifying voltage sag domains generated based on multiple uncertainty scenarios
The present disclosure discloses a system for identifying a voltage sag domain generated based on multiple uncertainty scenarios. The system includes a processor, a photovoltaic monitoring device, an electricity monitoring device, and regulation hardware. The processor is configured to: collect photovoltaic output power information of each photovoltaic site based on the photovoltaic monitoring device deployed at the each photovoltaic site on a distribution network, and train a DCGAN based on the photovoltaic output power information; generate a photovoltaic output scene using the DCGAN; determine, based on the photovoltaic output scene, one or more given parameters, a sensitive load threshold, and one or more structural parameters of an electricity system, a voltage sag domain identification result corresponding to the one or more given parameters and the sensitive load threshold.
1 . A system for identifying a voltage sag domain generated based on multiple uncertainty scenarios, wherein the system includes a processor, at least one photovoltaic monitoring device, an electricity monitoring device, and a regulation hardware;
the processor is configured to:
collect photovoltaic output power information of each photovoltaic site based on the photovoltaic monitoring device deployed at each photovoltaic site on a distribution network, and train a deep convolutional generative adversarial network based on the photovoltaic output power information;
generate a photovoltaic output scene using the deep convolutional generative adversarial network;
determine, based on the photovoltaic output scene, one or more given parameters, a sensitive load threshold, and one or more structural parameters of an electricity system, a voltage sag domain identification result corresponding to the one or more given parameters and the sensitive load threshold;
determine a voltage at each node of the distribution network before a fault based on the photovoltaic output scene using a Newton-Raphson algorithm;
obtain a current bus from a line of the distribution network, calculate an impedance matrix of each node of the distribution network by an additional branch manner or an admittance matrix inversion based on the one or more given parameters, and determine a voltage sag amplitude of a node where a sensitive load is located when different types of short-circuit faults occur at nodes on the current bus based on the voltage at each node before the fault and the impedance matrix of each node;
determine a voltage sag magnitude vector for a bus node based on the voltage sag amplitude of the node where the sensitive load is located when the different types of short-circuit faults occur at the nodes on the current bus;
obtain a difference vector by comparing the voltage sag magnitude vector of the bus node with the sensitive load threshold; and determine a line correlation vector based on the difference vector;
perform a judgment to determine whether the current bus satisfies an analysis condition based on the line correlation vector;
in response to determine that the current bus does not satisfy the analysis condition, determine that the current bus is not in the voltage sag domain and select a next line from the distribution network as the current bus to continue the judgment; or
in response to determine that the current bus satisfies the analysis condition, obtain, by analyzing the current bus using a Newton quadratic interpolation algorithm, a voltage sag amplitude analytical formula and a fault voltage equation for the current bus; and obtain a portion of the current bus within the voltage sag domain based on the fault voltage equation, and select a next line from the distribution network as the current bus to continue the judgment; and
when all lines of the distribution network are traversed and iterated, obtain a plurality of voltage sag domains corresponding to the one or more given parameters and the sensitive load threshold, wherein a concatenation of the plurality of voltage sag domains is the voltage sag domain identification result;
generate a voltage reduce response instruction based on the voltage sag domain identification result; the voltage reduce response instruction including a monitoring frequency, a monitoring accuracy, an upload interval period, and a warning threshold for indicating the electricity monitoring device;
send the voltage reduce response instruction to the electricity monitoring device of the voltage sag domain;
the electricity monitoring device is configured to:
collect an electricity parameter sequence in the distribution network at the monitoring frequency and with the monitoring accuracy, and upload the electricity parameter sequence to the processor at the upload interval period;
determine whether voltage data in the electricity parameter sequence is below the warning threshold; and
in response to determining that the voltage data in the electricity parameter sequence is below the warning threshold, send a warning signal to the processor; and
the processor is further configured to:
determine an estimated impact load corresponding to a node where the electricity monitoring device is located based on the warning signal; and
generate a regulation instruction and issue the regulation instruction to regulation hardware corresponding to the estimated impact load to control the regulation hardware to perform a compensation work.
2 . The system of claim 1 , wherein a property of the deep convolutional generative adversarial network includes: using a fully convolutional network, eliminating a fully connected layer, and using a batch normalization; and
the processor is further configured to:
after preprocessing the photovoltaic output power information, designate a random sampling noise as an input of a generator of the deep convolutional generative adversarial network, designate historical output power data of the each photovoltaic site as an input of a discriminator of the deep convolutional generative adversarial network, and train the deep convolutional generative adversarial network using training data until a Nash equilibrium is reached; and
when the training of the deep convolutional generative adversarial network is completed, obtain a photovoltaic output scene generated by the generator that is similar to real data.
3 . The system of claim 1 , wherein the processor is further configured to:
by substituting actual data, form the admittance matrix based on clarifying the one or more structural parameters of the electricity system, and calculate an offset of each of the one or more structural parameters based on given initial values; and
perform an error analysis, and in response to determining that an error satisfies an error precision, jump out of a loop and output a result; in response to determining that the error does not satisfy the error precision, continue to solve a Jacobi matrix and solve for a correction voltage by a correction equation, and then perform the error analysis again.
4 . The system of claim 1 , wherein the different types of short-circuit faults include a three-phase short-circuit fault, a single-phase grounded short-circuit fault, a two-phase short-circuit fault, and a two-phase grounded short-circuit fault;
when the short-circuit fault is the three-phase short-circuit fault, the processor is configured to:
when the three-phase short-circuit fault occurs at a fault point K, which indicates that the short-circuit fault is a symmetrically balanced fault, consider only a positive sequence, wherein three-phase voltage sag magnitudes at a sensitive load node S are the same magnitude, one of the three-phase voltage sag magnitudes is:
U
s
f
=
U
s
p
f
-
(
Z
S
K
1
Z
K
K
1
)
U
K
p
f
where,
U
s
f
denotes the voltage sag amplitude at the sensitive load node S,
Z
K
K
1
denotes a positive sequence self-impedance,
Z
S
K
1
denotes a positive sequence mutual impedance,
U
s
p
f
denotes a voltage before fault at the sensitive load node S,
U
K
p
f
denotes a voltage before fault at the fault point K; the voltage sag amplitude
U
s
f
at the fault point node S is related to the one or more structural parameters of the electricity system and an operation state of the electricity system before fault, and when an asymmetrical short-circuit fault whose phase is a special phase occurs at the fault point K, a symmetrical component manner is used for analysis;
when the short-circuit fault is the single-phase grounded short-circuit fault, the processor is configured to:
when the single-phase grounded short-circuit fault occurs at the fault point K with a phase A as a special phase, determine the three-phase voltage sag amplitudes at the sensitive load node S as follows:
{
U
A
,
S
f
=
U
A
,
S
pf
-
Z
SK
0
+
Z
SK
1
+
Z
SK
2
Z
KK
0
+
Z
KK
1
+
Z
KK
2
U
K
pf
U
B
,
S
f
=
α
2
U
B
,
S
pf
-
Z
SK
0
+
α
2
Z
SK
1
+
α
Z
SK
2
Z
KK
0
+
Z
KK
1
+
Z
KK
2
U
K
pf
U
C
,
S
f
=
α
U
C
,
S
pf
-
Z
SK
0
+
α
Z
SK
1
+
α
2
Z
SK
2
Z
KK
0
+
Z
KK
1
+
Z
KK
S
U
K
pf
where
U
A
,
S
f
denotes an A phase voltage sag amplitude at the sensitive load node
U
B
,
S
f
denotes a B phase voltage sag amplitude at the sensitive load node S,
U
C
,
S
f
denotes a C phase voltage sag amplitude at the sensitive load node S; and
U
A
,
S
p
f
denotes an A phase voltage before fault at the sensitive load node S,
U
B
,
S
p
f
denotes a B phase voltage before fault at the sensitive load node S,
U
C
,
S
pf
denotes a C phase voltage before fault at the sensitive load node S;
Z
KK
0
denotes a zero sequence self-impedance at the fault point K,
Z
KK
1
denotes a positive sequence self-impedance at the fault point K,
Z
KK
2
denotes a negative sequence self-impedance at the fault point K;
Z
SK
0
denotes a zero sequence mutual impedance between the sensitive load node S and the fault point K,
Z
SK
1
denotes a positive sequence mutual impedance between the sensitive load node S and the fault point K,
Z
SK
3
denotes a negative sequence mutual impedance between the sensitive load node S and the fault point K;
U
K
pf
denotes the voltage before fault at the fault point K; α denotes a rotation factor, α=e j120° ;
when the short-circuit fault is the two-phase short-circuit fault, the processor is configured to:
when the two-phase short-circuit fault occurs between the B phase and the C phase, determine the three-phase voltage sag amplitudes at the point as:
{
U
A
,
S
f
=
U
A
,
S
pf
-
Z
SK
1
-
Z
SK
2
Z
KK
1
+
Z
KK
2
U
K
pf
U
B
,
S
f
=
α
2
U
B
,
S
pf
-
α
2
Z
SK
1
-
α
Z
SK
2
Z
KK
1
+
Z
KK
2
U
K
pf
U
C
,
S
f
=
α
U
C
,
S
pf
-
α
Z
SK
1
+
α
2
Z
SK
2
Z
KK
1
+
Z
KK
S
U
K
pf
when the short-circuit fault is the two-phase grounded short-circuit fault, the processor is configured to:
when the two-phase grounded short-circuit fault occurs between the B phase and the C phase, determine the three-phase voltage sag amplitudes at the sensitive load node S as follows:
U
A
,
S
f
=
U
A
,
S
pf
-
(
Z
SK
1
-
Z
SK
0
)
Z
KK
2
+
(
Z
SK
1
+
Z
SK
2
)
Z
KK
0
Z
KK
0
Z
KK
1
+
Z
KK
1
Z
KK
2
+
Z
KK
2
Z
KK
0
U
K
pf
U
B
,
S
f
=
α
2
U
B
,
S
pf
-
(
α
2
Z
SK
1
-
Z
SK
0
)
Z
KK
+
(
α
2
Z
SK
1
-
α
Z
SK
2
)
Z
KK
0
Z
KK
0
Z
KK
1
+
Z
KK
1
Z
KK
2
+
Z
KK
2
Z
KK
0
U
K
pf
U
C
,
S
f
=
α
U
C
,
S
pf
-
(
α
Z
SK
1
-
Z
SK
0
)
Z
KK
2
+
α
Z
SK
1
+
α
2
Z
SK
)
Z
KK
0
Z
KK
0
Z
KK
1
+
Z
KK
1
Z
KK
2
+
Z
KK
2
Z
KK
0
U
K
pf
when the two-phase grounded short-circuit fault occurs on the bus of the electricity system, directly call self-impedances and mutual impedances from an impedance matrix of each node of the distribution network; however, when the
short-circuit fault occurs at a certain point on the line, calculate the self-impedances and the mutual impedances by introducing a position variable p(0≤p≤1) to juggle a node impedance matrix, wherein the sensitive load node
S denotes a bus node where the sensitive load is located, Z C denotes a sequence impedance of a line F−T, when the fault point K is moving on the line F−T, three sequence self-impedance
Z
KK
i
(
i
=
0
,
1
,
2
)
and three sequence mutual impedance
Z
SK
i
between the fault point K and the sensitive load node S are both expressed by the position variable p of the impedance matrix Z:
Z
KK
0
1
2
=
p
2
(
Z
FF
0
1
2
+
Z
TT
0
1
2
-
2
Z
FT
0
1
2
-
Z
C
0
1
2
)
+
p
[
Z
C
0
1
2
-
2
(
Z
FF
0
1
2
-
Z
FT
0
1
2
)
]
+
Z
FF
0
1
2
Z
SK
0
1
2
=
Z
SF
0
1
2
+
p
(
Z
ST
0
1
2
-
Z
SF
0
1
2
)
where,
Z
FF
0
1
2
and
Z
TT
0
1
2
respectively denote the zero sequence self-impedance, the positive sequence self-impedance, and the negative sequence self-impedance of system bus nodes F and T,
T
,
Z
FT
0
1
2
,
Z
SF
0
1
2
,
and
Z
S
T
0
1
2
respectively denote a sequence mutual impedance of the system bus nodes F and T, a sequence mutual impedance between the bus node F and the sensitive load node S, a sequence mutual impedance between the bus node T and the sensitive load node S, which are all called from the impedance matrix of each node of the distribution network,
Z
C
0
1
2
denotes a line sequence impedance between the nodes F and T, and p(0≤p≤1) denotes a position variable of the fault point K on the line F−T;
a voltage
U
K
pf
at the fault point K before fault is represented by a fault position variable p:
U
F
pf
=
U
T
pf
+
p
(
U
T
pf
-
U
F
pf
)
where,
U
F
pf
denotes the voltage at the bus node F before fault,
U
T
pf
denotes the voltage at the bus node T before fault, p(0≤p≤1) denotes the position variable of the fault point K on the line F−T; the voltage sag amplitude at the sensitive load node S is expressed by the voltage before fault and the each sequence impedance, to obtain a function expression U(p) of the voltage sag amplitude at the sensitive load node S related to the position variable p when any short-circuit fault occurs at any fault point.
5 . The system of claim 1 , wherein an expression of the difference vector is:
Δ
U
s
=
[
Δ
U
S
,
1
Δ
U
S
,
2
⋮
Δ
U
S
,
n
]
=
[
❘
"\[LeftBracketingBar]"
U
S
_
1
f
❘
"\[RightBracketingBar]"
❘
"\[LeftBracketingBar]"
U
S
_
2
f
❘
"\[RightBracketingBar]"
⋮
❘
"\[LeftBracketingBar]"
U
S
_
n
f
❘
"\[RightBracketingBar]"
]
-
[
U
th
U
th
⋮
U
th
]
where,
❘
"\[LeftBracketingBar]"
U
S
_
1
f
❘
"\[RightBracketingBar]"
denotes an absolute value of a voltage amplitude at the sensitive load node S when the short-circuit fault occurs in a bus 1,
❘
"\[LeftBracketingBar]"
U
S
_
2
f
❘
"\[RightBracketingBar]"
denotes an absolute value of a voltage amplitude at the sensitive load node S is located when the short-circuit fault occurs in a bus 2,
❘
"\[LeftBracketingBar]"
U
S
_
n
f
❘
"\[RightBracketingBar]"
denotes an absolute value of a voltage amplitude at the node S of where the sensitive load is located when the short-circuit fault occurs in a bus n, and U th denotes a sensitive load voltage sag threshold; ΔU S,1 denotes a difference between the absolute value of the voltage amplitude at the node S of where the sensitive load is located when the short-circuit fault occurs in the bus 1 and the sensitive load voltage sag threshold, ΔU S,1 denotes a difference between the absolute value of the voltage amplitude at the node S of where the sensitive load is located when the short-circuit fault occurs in the bus 2 and the sensitive load voltage sag threshold; ΔU S,n denotes a difference between the absolute value of the voltage amplitude at the node S of where the sensitive load is located when the short-circuit fault occurs in the bus n and the sensitive load voltage sag threshold and, ΔU S denotes the difference vector formed by ΔU S,1 , ΔU S,2 , and ΔU S,n ;
by determining positive and negative of elements in the difference vector ΔU S , whether the each node of the distribution network is within a voltage sag domain of the sensitive node is obtained, and a determination vector B of the node is introduced:
B
=
[
B
1
B
2
⋮
B
n
]
,
B
i
=
{
1
,
Δ
U
s
,
i
≤
0
0
,
Δ
U
s
,
i
>
0
where, B 1 denotes a determination result of the bus 1, B 2 denotes a determination result of the bus 2, Bn denotes a determination result of bus n, B i denotes the determination result of a bus i, and B denotes the node determination vector; ΔU S,i denotes a difference between the absolute value of the voltage amplitude at the node S of where the sensitive load is located when the short-circuit fault occurs in the bus i and the sensitive load voltage sag threshold;
B i =1 indicates that the bus i is within the voltage sag domain corresponding to the sensitive load; B i =0 indicates that the bus i is outside the voltage sag domain corresponding to the sensitive load, and an inclusion situation of each line in the voltage sag domain is determined by introducing a line correlation vector L:
L
=
[
L
1
L
2
⋮
L
m
]
=
[
B
1
_
F
+
B
1
_
T
B
2
_
F
+
B
2
_
T
⋮
B
m
_
F
+
B
m
_
T
]
where, L denotes the line correlation vector, L 1 denotes a determination result of the inclusion situation where a line 1 is included in the sag domain, L 2 denotes a determination result of the inclusion situation where a line 2 is included in the sag domain, and L m denotes a determination result of the inclusion situation where a line m is included in the sag domain; B 1_F denotes a sag domain determination result of the bus node F of the line 1, B 1_T denotes a sag domain determination result of the bus node T of the line 1, B 2_T denotes the sag domain determination result of the bus node T of the line 2, B m_F denotes the sag domain determination result of the bus node F of the line m, and B m_T denotes the sag domain determination result of the bus node T of the line m;
the perform a judgment to determine whether the current bus satisfies an analysis condition based on the line correlation vector, includes that:
if L i =0, it indicates that the line is not in the voltage sag domain, a next line is calculated directly; if L i =1, it indicates that one of a first node and a last node of the line is in the voltage sag domain, then it is obtained that the line is partially located in the sag domain, and there is a unique critical point on the line; if L i =2, it indicates that the first node and the last node of the line are both in the sag domain, a golden section search manner is used to solve a maximum value U max of a voltage sag amplitude curve, and the maximum value U max is compared with the sensitive load voltage sag threshold U th ; if U max <U th , it indicates that the line is completely within the sag domain, and there is no critical point, then the next line is directly calculated; if U max ≥U th , it indicates that the line i is partially located in the sag domain, and there are two critical points on the line;
an entire process of the golden section search manner includes: finding a point P max in p a ≤p<p b (p a =0,p b =1), which is able to maximize a sag voltage of a point of common coupling, and using P max as an interpolation point to ensure that an interpolation curve passes through a maximum point of the voltage sag amplitude, and providing an initial value point for a positive cut iteration.
6 . The system of claim 1 , wherein the processor is configured to:
designate a root P ia of the fault voltage equation as an initial iteration value of a positive cut iteration manner, obtain a precise critical point position by iteration, wherein an iteration expression of a critical fault distance obtained based on the positive cut iteration manner is as follows:
P
k
+
1
=
P
k
-
(
U
(
p
k
)
-
U
t
h
)
(
p
k
-
p
k
-
1
)
U
(
p
k
)
-
U
(
p
k
-
1
)
where, P k+1 denotes the critical fault distance for the (k−1)th solution, p k denotes the critical fault distance for the kth solution, p k+1 denotes the critical fault distance solved for the (k+1)th solution, U(p k ) denotes the voltage sag amplitude at the node S whose position variable is p k , U(p k−1 ) denotes the voltage sag amplitude at the node S whose position variable is p k−1 , and U th denotes a sensitive load voltage sag threshold; and
a convergence condition for the positive cut iteration manner is:
❘
"\[LeftBracketingBar]"
❘
"\[LeftBracketingBar]"
U
(
p
k
+
1
)
❘
"\[RightBracketingBar]"
-
U
t
h
❘
"\[RightBracketingBar]"
<
ε
where, U(p k−1 ) denotes the voltage sag amplitude at the node S whose position variable is p k−1 , U th denotes the sensitive load voltage sag threshold.
7 . The system of claim 1 , wherein the regulation hardware includes one or more of a backup electricity supply, a dynamic voltage regulator, and a load switch; and the regulation instruction includes one or more of a backup electricity supply instruction, a voltage self-regulation instruction, and a switch instruction;
the backup electricity supply is configured to: provide an electricity supply corresponding to the estimated impact load after receiving the backup electricity supply instruction;
the dynamic voltage regulator is configured to: provide a compensating voltage corresponding to the estimated impact load after receiving the voltage self-regulation instruction; and
the load switch is configured to: control the estimated impact load and connection and disconnection of the distribution network after receiving the switch instruction.
8 . The system of claim 1 , wherein the processor is further configured to:
construct an electricity feature map based on the one or more given parameters, the electricity parameter sequence, the photovoltaic output power information, line base data, future lighting data, and a sensitive load threshold;
obtain, by inputting the electricity feature map into a voltage sag risk model, a voltage sag risk region; and
determine, by calculating a plurality of voltage sag domains within the voltage sag risk region corresponding to the one or more given parameters and the sensitive load threshold, the voltage sag domain identification result.
9 . The system of claim 8 , wherein a training process of the voltage sag risk model includes an initial training phase and an intensive training phase performed separately using training data; wherein the initial training phase is performed by constructing the training data based on a generalized dataset obtained from a cloud platform; the intensive training phase is performed by constructing the training data based on data actually collected from the distribution network;
the training data includes training samples and labels corresponding to the training samples; the training samples including a sample electricity feature map constructed based on one or more sample given parameters, a sample electricity parameter sequence, sample photovoltaic output power information, sample line base data, sample future lighting data, and a sample sensitive load threshold; the labels corresponding to the training samples being a region of the voltage sag domain corresponding to the sample sensitive load;
sample sizes corresponding to the distribution network of different complexity levels in the training data of the initial training phase is not less than a preset threshold, the preset threshold being determined based on the complexity level of the corresponding electricity grid.